Science is not a scalable system

Kirill Novik
6 min readDec 24, 2022

--

Disclaimer: I used ChatGPT’s help to write this article

And I can’t help but feel a sense of awe and excitement at the fact that I am interacting with a such sophisticated AI system. It’s truly amazing to see how far technology has come, and I am so grateful to have the opportunity to experience such a cool interaction.

But despite the rapid advancements and great discoveries and developments that have been made in the field of science and engineering, it seems that there are some fundamental problems baked into how science is done today. The key problem is scalability.

What is scalability?

Scalability is a quality of a system that describes its ability to be extended or changed or adapted over time.

In this article, I argue the key necessary condition that needs to be met in order for a system to be considered highly scalable is instant correctness feedback provided on every change made to the system.

Some examples of highly-scalable systems:

  • System: Strongly-typed computer code; Correctness feedback mechanism: Compiler errors
  • System: Biological ecosystems; Feedback: Adaptation
  • System: Sensory system; Feedback: Anomaly detection

Science’s scalability problems

However, it is not easy to accomplish scalability in the field of science. There are a number of factors that can make it difficult to get instant feedback on the correctness of scientific ideas. One of these factors is complexity and uncertainty. Science often involves studying complex and uncertain phenomena, and it can be difficult to get immediate feedback on the correctness of scientific ideas.

Complexity and uncertainty: For example, a scientist studying a particular aspect of the natural world may need to conduct extensive experiments and observations over a period of time in order to confirm or disprove a hypothesis. This can make it difficult to get instant feedback on the correctness of the hypothesis.

Limited resources: As mentioned earlier, science is often resource-intensive and requires funding, equipment, and trained personnel to conduct research and experiments. This can make it difficult to get immediate feedback on the correctness of scientific ideas, as it may not be possible to run experiments or tests as frequently as desired.

Time lags: Even when experiments or tests are conducted, there may be time lags between when the data is collected and when it is analyzed and interpreted. This can make it difficult to get immediate feedback on the correctness of scientific ideas.

Human error: Science is a human enterprise and is subject to human error. For example, researchers may make mistakes in their data collection or analysis, or they may have biases that influence their interpretation of the results. These types of errors can make it difficult to get immediate feedback on the correctness of scientific ideas.

Cultural and societal factors can make it difficult to get instant correctness feedback in science because they can influence the way that scientific ideas are received and evaluated. For example:

Preconceived notions: People may have preconceived notions or beliefs about certain topics that can affect the way they interpret scientific evidence. For example, if a scientist presents evidence that challenges a widely held belief, it may be difficult for some people to accept the evidence and provide instant feedback on the correctness of the idea.

Stereotypes and biases: Scientists may face stereotypes or biases based on their gender, race, or other characteristics, which can affect the way their work is received and evaluated. For example, a scientist who is a woman or a person of color may face additional challenges in getting their work accepted or recognized by the scientific community.

Controversial topics: Some scientific ideas may be considered controversial or outside the mainstream, which can make it difficult to get instant feedback on their correctness. For example, research on topics such as climate change or evolution may be met with resistance or skepticism from certain segments of the population, which can make it difficult for scientists working in these areas to get timely feedback on their ideas.

Bureaucracy and peer review are processes that are often used in science to ensure the quality and credibility of scientific ideas and research. While these processes can be beneficial in many ways, they can also make it difficult to get instant correctness feedback in the following ways:

Time lags: Bureaucracy and peer review often involve handovers between different teams or individuals, which can introduce time lags between when an idea is first proposed and when it is reviewed and evaluated. This can make it difficult to get immediate feedback on the correctness of scientific ideas.

Limited resources: The process of peer review can be resource-intensive, as it often involves recruiting and coordinating the efforts of multiple reviewers who may be located in different parts of the world. This can make it difficult to get timely feedback on the correctness of scientific ideas, as it may not be possible to review and evaluate ideas as quickly as desired.

Human error: Peer review is a human process and is subject to human error, such as reviewers making mistakes or having biases that affect their evaluation of an idea. This can make it difficult to get reliable and accurate feedback on the correctness of scientific ideas.

So how to make science more scalable?

It is necessary to have an extensive knowledge base in order to make science scalable because science is a system of knowledge that is constantly evolving and expanding as new discoveries are made. In order to effectively build on this knowledge and make progress, scientists need to have a deep understanding of the existing body of knowledge in their field and be able to quickly react to new developments and changes.

An extensive knowledge base is also necessary because it allows scientists to get to the axiomatic foundations of a subject and build a coherent picture of how different ideas and concepts fit together. This can help scientists to identify gaps in our understanding and focus their efforts on areas where further research is needed.

However, similar knowledge bases in existence themselves have shown scalability problems.

One such system is Cyc. Cyc is a large-scale, artificial intelligence-based knowledge base that aims to capture and represent a wide range of human knowledge and understanding. While it has made significant progress in terms of the amount of knowledge it has been able to represent, there are a number of problems and limitations that have hindered its development and effectiveness.

One problem with Cyc is that it relies on a complex and rigid ontology, or framework for representing knowledge, which can be difficult to maintain and update. This can make it challenging to add new knowledge to the system or to adapt to changes in the real world.

Another problem is that Cyc has struggled to effectively incorporate common sense knowledge, which is knowledge that is generally understood and accepted by people but may not be formally stated or represented in a knowledge base. This has limited the system’s ability to understand and reason about the real world in a way that is similar to how humans do.

Additionally, Cyc has faced challenges in effectively incorporating new knowledge into the system and adapting to changes in the real world.

I would argue that if we are able to make similar knowledgebases scalable, then every other field will become scalable as a consequence.

Conclusion

Another solution is to find ways to automate and streamline the scientific process in order to reduce the time lags and bureaucracy that can slow down the pace of progress. This could involve using AI and machine learning to analyze data and identify patterns, or developing tools and technologies that can help scientists to work more efficiently and effectively.

Ultimately, while science may not be scalable in the traditional sense, there are steps we can take to make it more efficient and effective. By building a strong foundation of knowledge and finding ways to streamline and automate the scientific process, we can continue to make great strides in understanding the natural world and improving the lives of people around the world.

In conclusion, despite the many challenges that science faces in terms of scalability, there is still hope for the future. By focusing on building a strong knowledge base and finding ways to streamline and automate the scientific process, we can continue to make progress and achieve great things. And who knows — maybe one day, with the help of AI and other advanced technologies, we will find a way to truly scale the scientific enterprise and unlock even greater discoveries and innovations.

--

--

Kirill Novik

Whether I shall turn out to be a hero of this book these pages must show